Explore the latest Microsoft Applied Skill PL 7008: Create agents in Microsoft Copilot Studio

Introduction to Copilot Studio

Microsoft Copilot Studio is a visual, low-code platform designed for building and managing AI agents. An agent is an AI-driven conversational interface powered by large language models (LLMs) and additional knowledge sources. These intelligent agents can handle a wide range of tasks, from answering basic questions to resolving more complex issues that require in-depth conversations. Agents created with Copilot Studio can interact with customers and employees in multiple languages across various platforms, including websites, mobile apps, Facebook, Microsoft Teams, or any channel supported by the Azure Bot Service.

Possible use cases for Microsoft Copilot Studio include:

• Assisting with sales and support inquiries
• Providing store hours and other business information
• Addressing employee health and vacation benefits
• Offering public health tracking information
• Answering common employee questions in businesses

Copilot Studio allows users to easily create agents without the need for data scientists or developers. The Applied skill course is divided into 6 learning paths as given below.

Learning path 1 : Build an initial agent with Microsoft Copilot Studio

In this learning path, different methods used to create AI powered conversational agents is defined. You can create agent using the conversational builder, where in we can create the conversational flow using the drag and drop interface available in the studio. With the generative answers feature in Microsoft Copilot Studio, your agent can automatically retrieve and present information from various internal or external sources, eliminating the need to create specific topics. These answers can either serve as the primary source of information for your agent or act as a fallback when predefined topics cannot address a user’s query. This feature enables the rapid creation and deployment of a functional agent without the need to manually author numerous topics that may not fully meet customer needs. We can add public website URLs as knowledge source. It allows us to add some document as knowledge source as well.
What’s particularly important—and why this feature is being introduced first—is to encourage creators to think about available data sources that could be used to summarize answers for potential customer or employee questions. By identifying these data sources and integrating them into Copilot Studio, you may be able to focus on authoring a smaller set of topics through the authoring canvas, allowing you to deliver value to your customers or employees more quickly. This module also gives the description of how to publish the agent and deploy the same into various channels.

Learning path 2: Manage topics in Microsoft Copilot Studio

In Microsoft Copilot Studio, Topics define the paths customers follow during interactions with an agent. Topics are triggered by specific trigger phrases (e.g., “What are the store hours”) entered by users, which prompt the agent to ask questions, provide information, or take actions. These interactions are guided by conversation nodes, which define the agent’s responses, such as sending a message, asking for information, or directing the user to another topic.
Trigger phrases are key phrases that activate a topic and can be tailored to ensure accuracy, with multiple phrases per topic. Conversation nodes include options like asking questions, sending messages, managing variables, and calling external actions via Power Automate Flows. Nodes can also involve asking specific questions (e.g., capturing a city name) and handling responses with behavior settings like retries or skips. Messages, images, videos, and adaptive cards can be added to enhance the user experience. Quick replies offer predefined responses for smoother interactions. The Topic Management Node helps route users to different topics, while the End the Conversation node concludes the interaction, potentially triggering surveys or escalating to a live agent.
Overall, Copilot Studio allows for flexible and dynamic conversation flows, leveraging triggers, nodes, and external integrations to create personalized, efficient customer interactions.

Learning path 3: Work with entities and variables in Microsoft Copilot Studio

The agent can capture and store user inputs, such as a name or date of birth, for use in other parts of the interaction.
To enhance the user experience, Microsoft Copilot Studio offers two key features:

• Entities: These represent real-world data units, such as phone numbers, cities, zip codes, or a person’s name.
• Variables: These are pieces of information captured and stored within the agent for later use.

We can work with prebuilt entities which is already built in the copilot studio. We can also add our own custom entities which can be used by the agent. Variables help us to reuse it at different conversational engagements. We can reuse them across various topics.

Learning path 4: Enhance Microsoft Copilot Studio agents

Often, when an agent interacts with customers, it needs information from other applications to provide a personalized experience. For example, if a customer asks about the weather in their location, the agent can pass details like the customer’s city and postal code to a weather service. The service can then return the weather forecast, which the agent shares with the customer. In Microsoft Copilot Studio, these tasks are called actions. Actions trigger Power Automate flows to automate tasks, interact with operating systems, or connect with external applications. For instance, an action can pass location data to an external service and send the weather details back to the agent.

Power Automate flows are triggered within topics using the “Call an action” node. You can use an existing flow from your Power Apps environment or create one within the Copilot Studio authoring canvas. To enable interaction with an agent, a special Microsoft Copilot Studio trigger is required. This trigger captures data from the agent and sends response information back.

Power Automate offers the following:

• Microsoft Copilot Studio Trigger: Specifies input parameters captured from the agent.
• Return Value(s) to Microsoft Copilot Studio Action: Defines output parameters sent back to the agent.

When you create a flow from the Copilot Studio agent, a starter template is loaded with the trigger and return values action. You’ll need to define input and output parameters and complete the flow structure. Once your flow is created, you can trigger it from the agent’s topic using the “Call an action” node. Variables can be passed as input parameters to the flow. Create a topic with trigger phrases like “Will it rain?” or “What’s the weather?” to start. To pass location information, use the “Ask a question” node to capture details such as city and zip code. These responses are stored in variables. After defining the questions, add a “Call an action” node and select the weather flow you created earlier.

Map the input parameters in the flow to the corresponding output variables from the question nodes. For example, the “City” (text) value maps to the “City” (city) variable, and “Zipcode” (number) maps to “Zip” (number). To transfer conversations from an agent to a customer engagement hub, such as Dynamics 365 Customer Service, you must configure the agent in Microsoft Copilot Studio. The agent can be connected to various hubs, including Dynamics 365, Genesys, Salesforce, and others. Ensure the agent is published before validating the integration.

The configuration also automatically enables a Teams channel for communication between Copilot Studio and Dynamics 365. Once set up, you can continue the configuration in Omnichannel for Customer Service.

Learning path 5 : Create an agent with Microsoft Copilot Studio and Dataverse for Teams

To create agents in Microsoft Copilot Studio, first install the app in Microsoft Teams. This app allows you to create, edit, test, publish agents, and monitor their performance. Team owners and the agent creator will have full access, while team members will only have read access. If it’s your first time creating an agent, you will need to wait for a notification confirming that the setup is complete. Then, name the agent (e.g., Event Contacts), select the language (which cannot be changed later), and optionally choose an icon. Finally, click “Create” to finish setting up the agent. Once created, you can manage its topics, test its performance, and publish it using the Copilot Studio editor.

Learning path 6 : Guided Project

Here, a project scenario is given in such a way that, the learners can create an end-to-end project on agent creation. The following exercises are provided as part of this project as below.

• Create a custom table
• Build an agent in Microsoft Copilot Studio with the new AI capabilities
• Create and manage topics
• Set up Generative AI

Overall, this applied skill helps the candidates to create an agent from scratch and build on various features to it, in such a way that an end-to-end Copilot agent with capabilities for conversation and action can be leveraged.

Reference:  Create agents in Microsoft Copilot Studio – Training | Microsoft Learn

About CloudThat

CloudThat is a leading provider of Cloud Training and Consulting services with a global presence in India, the USA, Asia, Europe, and Africa. Specializing in AWS, Microsoft Azure, GCP, VMware, Databricks, and more, the company serves mid-market and enterprise clients, offering comprehensive expertise in Cloud Migration, Data Platforms, DevOps, IoT, AI/ML, and more.

CloudThat is the first Indian Company to win the prestigious Microsoft Partner 2024 Award and is recognized as a top-tier partner with AWS and Microsoft, including the prestigious ‘Think Big’ partner award from AWS and the Microsoft Superstars FY 2023 award in Asia & India. Having trained 650k+ professionals in 500+ cloud certifications and completed 300+ consulting projects globally, CloudThat is an official AWS Advanced Consulting Partner, Microsoft Gold Partner, AWS Training Partner, AWS Migration Partner, AWS Data and Analytics Partner, AWS DevOps Competency Partner, AWS GenAI Competency Partner, Amazon QuickSight Service Delivery Partner, Amazon EKS Service Delivery Partner,  AWS Microsoft Workload Partners, Amazon EC2 Service Delivery Partner, Amazon ECS Service Delivery Partner, AWS Glue Service Delivery Partner, Amazon Redshift Service Delivery Partner, AWS Control Tower Service Delivery Partner, AWS WAF Service Delivery Partner and many more.

To get started, go through our Consultancy page and Managed Services Package, CloudThat’s offerings.